Per-message transparency

Every assistant reply carries a small shield badge. Click it to open the Governance drawer for that reply — a plain-language summary of how the answer was produced and what protected it.

Trust & Control

The drawer opens on Trust & Control, headed by an overall status — Protected · All governance controls active when everything ran as expected — followed by six rows:

Row	What it tells you
Assistant	Which assistant answered — for example, SW Business Assistant
Safety	That input and output were both checked
Your data	That your data stayed org-only, isolated to your account
Knowledge	Which sources were used, or No external sources if the reply drew on none
Boundaries	That no actions were taken without your approval
Retention	How long this is kept — for example, 90 days

Below those, This turn summarizes what happened on this specific exchange.

You can pin the drawer open while you keep working, expand it for a wider view, or close it.

It describes the reply, not just the assistant

The same assistant can produce differently-governed answers depending on what you asked and what it drew on — so the drawer reflects this reply. Open it on a different message and the Knowledge row, in particular, may read differently.